LC-SVC: A Low-Complexity Sparse Vector Coding Scheme for Reliable Short-Packet Transmission in MIMO-Based Wireless...

Reliable short-packet transmission is essential for MIMO-based wireless sensor systems, where low-power terminals upload short data blocks to multi-antenna gateways under stringent latency and complexity constraints. Although multi-antenna combining improves the equivalent received signal-to-noise ratio, the post-combining recovery of sparse-vector-coded packets is still limited by the high inter-column correlation of short binary spreading matrices and redundant searches over invalid indices in conventional orthogonal matching pursuit. To address these issues, this paper proposes a low-complexity sparse vector coding scheme for reliable short-packet transmission in MIMO-based wireless sensor systems. Specifically, at the transmitter, a low-coherence binary spreading matrix (LCB-SM) construction algorithm is designed based on Hadamard initialization and column-wise correlation optimization, which improves the distinguishability of sparse support positions while preserving multiplication-free encoding. At the receiver, a frozen-index-pruned orthogonal (FIP-OMP) matching pursuit algorithm is developed to exploit the predefined sparse mapping rule, thereby excluding invalid indices during atom selection and reducing noise-induced false support detection. Simulation results show that, for N=24, Ls=16, and K=2, the LCB-SM achieves a signal-to-noise (SNR) ratio gain of approximately 1.2 dB over the optimized partial hadamard matrix (OPHM) at a BLER of 10-2; for K=3, it reduces the high-SNR BLER by more than two orders of magnitude compared with OPHM. Meanwhile, the FIP-OMP achieves an approximately 0.5-0.8 dB gain over conventional orthogonal matching pursuit at a BLER of 10-2 with an average decoding time close to that of orthogonal matching pursuit (OMP) and much lower than that of multipath matching pursuit, Dynamic OMP and compressive sampling matching pursuit. A software defined radio-based wireless experiment further validates the reconstruction capability of LC-SVC for continuous sensing data, achieving an RMSE of 0.0612. These results demonstrate that LC-SVC improves the reliability of the considered short-packet recovery task with lightweight transmitter-side encoding and low gateway-side decoding overhead.